Heterogeneous Sensor Data Fusion for Target Classification Using Adaptive Distance Function

2021-04-01
Atıcı, Bengü
Karasakal, Esra
Karasakal, Orhan
Automatic Target Recognition (ATR) systems are used as decisionsupport systems to classify the potential targets in military applications. Thesesystems are composed of four phases, which are selection of sensors, preprocessingof radar data, feature extraction and selection, and processing of features toclassify potential targets. In this study, the classification phase of an ATR systemhaving heterogeneous sensors is considered. We propose novel multiple criteriaclassification methods based on the modified Dempster–Shafer theory. Ensembleof classifiers is used as the first step probabilistic classification algorithm. Artificialneural network and support vector machine are employed in the ensemble. Eachnon-imaginary dataset coming from heterogeneous sensors is classified by bothclassifiers in the ensemble, and the classification result that has a higher accuracyratio is chosen for each of the sensors. The proposed data fusion algorithms areused to combine the sensors’ results to reach the final class of the target.We presentextensive computational results that show the merits of the proposed algorithms.
Citation Formats
B. Atıcı, E. Karasakal, and O. Karasakal, Heterogeneous Sensor Data Fusion for Target Classification Using Adaptive Distance Function. 2021.